Revue de presse générée automatiquement avec
Profil généré automatiquement
Pulseon
Un individu à l'aise avec la technologie, fasciné par la robotique et l'intelligence artificielle, avec un fort intérêt pour les progrès technologiques et les innovations émergentes qui façonnent l'avenir. Il recherche des mises à jour approfondies sur les derniers développements et les avancées dans ces domaines.
Robotics (30%)Artificial Intelligence (30%)Technological Advancements (20%)Innovation and Emerging Technologies (20%)
Vous souhaitez recevoir chaque jour la revue de presse de ce profil ?
Underwater robotics, Gemini 3, AI funding, space data centers...
Jeudi 18 décembre 2025 à 06:50
Robotics & Autonomous Systems
Skana Robotics enables underwater fleets to “talk” to each other
Skana Robotics has rolled out a decision‑making algorithm that lets unmanned underwater vessels share sensor data and coordinate actions in real time, dramatically improving swarm navigation and mission efficiency. The system lets each robot dynamically adapt to the movements of its peers, a breakthrough for oceanic exploration and subsea infrastructure inspection.
TechCrunch
Unitree launches the world’s first robot app store
Unitree’s new “Robot App Store” offers plug‑and‑play modules ranging from simple locomotion tricks to advanced perception stacks, letting developers upload and download canned actions for their bots with a few clicks. The marketplace mirrors consumer‑grade app ecosystems and could accelerate the commercialization of service robots across logistics, hospitality, and education.
Tech Radar
UMA bets on European humanoid and mobile robots
Paris‑based startup UMA (Universal Mechanical Assistant) emerged from stealth with a team of ex‑Tesla, DeepMind and Hugging Face veterans, aiming to produce general‑purpose mobile and humanoid robots for industrial pilots by 2026. Backed by investors such as Greycroft and AI luminaries Yann LeCun, UMA plans dual platforms: a dual‑arm industrial unit for warehouses and a compact humanoid for human‑centric environments.
FrenchTechJournal
FrenchTechJournal
Artificial Intelligence & Machine Learning
Google’s Gemini 3 Flash raises the bar for speed and reasoning
Google unveiled Gemini 3 Flash, a slimmer version of its Gemini 3 family that delivers faster inference and higher scores on benchmarks like GPQA Diamond and Humanity’s Last Exam. The model is now the default in the Gemini app and will be accessible via the Gemini API, Vertex AI, and the new Antigravity IDE, positioning it for rapid adoption in search and developer tools.
Ars Technica
TechCrunch
OpenAI’s funding surge and Amazon’s $10 billion AI‑chip partnership
OpenAI secured a $1 billion investment from Disney and is in advanced talks for a $10 billion infusion from Amazon, which would see the AI lab adopt Amazon’s Trainium chips and AWS cloud services. The dual cash flow strengthens OpenAI’s valuation, now approaching $750 billion, while cementing Amazon’s role as a key hardware supplier in the generative‑AI ecosystem.
TechCrunch
The Information
AI‑generated code introduces more defects, study finds
An analysis of 470 open‑source pull requests shows that code produced by AI coding assistants carries significantly more bugs across logic, security, performance and maintainability categories than human‑written code. The findings echo concerns raised by SD Times about hallucinations and insecure suggestions from AI‑driven development tools.
DevOps.com
SD Times
AI permeates every stage of software development in 2025
The SD Times year‑in‑review reports that AI copilots now appear across the entire software lifecycle—from code generation and automated testing to vulnerability detection and observability—transforming developer workflows but also surfacing new challenges around hallucinations and security. Companies such as Google, GitHub and Microsoft have integrated “agent mode” and “Copilot Autofix” features to streamline multi‑file edits and real‑time debugging.
SD Times
Technological Advancements
Intel adds experimental AI‑assisted video‑encode APIs to libvpl
Intel’s latest libvpl 2.16 release introduces experimental interfaces that let developers harness AI models for smarter video compression, promising higher visual quality at lower bitrates for streaming and broadcast applications. The move signals a broader industry push to embed generative‑AI techniques into traditional media pipelines.
Phoronix
Rapidus pursues glass‑substrate panel‑level packaging for next‑gen AI chips
At SEMICON Japan, Rapidus showcased early work on panel‑level packaging using 600 × 600 mm glass substrates, a technique that could enable ultra‑dense chiplet integration for AI and high‑performance computing processors. The aggressive roadmap aims to leapfrog rivals by combining glass‑core substrates with advanced interconnects.
Tom's Hardware
Innovation & Emerging Technologies
Billionaires eye orbital data centers to power AI workloads
Tech moguls are proposing massive AI‑focused data centers in Earth orbit, citing the soaring power, water and land demands of terrestrial AI farms. Six proposals in 2025 envision gigawatt‑scale facilities that could mitigate terrestrial environmental impacts while delivering low‑latency compute for satellite‑based services.
The Verge
OpenAI launches a public ChatGPT App Directory and SDK
OpenAI opened a searchable App Directory that lets developers submit interactive “apps” that run inside ChatGPT’s UI, expanding the platform with tools ranging from productivity assistants to real‑time data connectors. The move marks a shift toward a modular AI ecosystem where third‑party functionality can be discovered and installed like traditional mobile apps.
The Verge
Aller aux sources
14 sources citées